Enzymatic Ligation Creates Discrete Multinanoparticle Building Blocks

Jun 28, 2008 - Christine M. Micheel, Jean M. J. Fréchet,* and A. Paul Alivisatos*. Department of Chemistry, UniVersity of California, Berkeley, Calif...
1 downloads 0 Views 1MB Size
Published on Web 06/28/2008

Enzymatic Ligation Creates Discrete Multinanoparticle Building Blocks for Self-Assembly Shelley A. Claridge, Alexander J. Mastroianni, Yeung B. Au, Huiyang W. Liang, Christine M. Micheel, Jean M. J. Fre´chet,* and A. Paul Alivisatos* Department of Chemistry, UniVersity of California, Berkeley, California 94720-1460, and DiVision of Materials Science, Lawrence Berkeley National Laboratory, Berkeley, California 94720 Received April 11, 2008; E-mail: [email protected]; [email protected]

Abstract: Enzymatic ligation of discrete nanoparticle-DNA conjugates creates nanoparticle dimer and trimer structures in which the nanoparticles are linked by single-stranded DNA, rather than by double-stranded DNA as in previous experiments. Ligation was verified by agarose gel and small-angle X-ray scattering. This capability was utilized in two ways: first, to create a new class of multiparticle building blocks for nanoscale self-assembly and, second, to develop a system that can amplify a population of discrete nanoparticle assemblies.

Introduction

Recent research in nanoassembly has demonstrated that DNA may be used to control the self-assembly of inorganic nanoparticles into designed functional materials1–3 by exploiting sequence-specific base pairing that governs its native assembly. In living systems, DNA serves as genetic information, a function so vital that nature has developed a multitude of enzymes to control its replication, correct errors in DNA sequences, and even destroy foreign DNA that enters a cell.4 Certain DNAmanipulating enzymes, such as restriction enzymes5,6 and polymerases,7 also have proven to be active on DNA-nanoparticle conjugates, suggesting that it may ultimately be possible to leverage the library of DNA-active enzymes to create more complex self-assembled nanomaterials and perhaps even to enable such functions as self-replication and error correction in a materials context. Nature’s enzymatic toolkit is an important and underutilized asset in DNA-mediated self-assembly of functional materials. Here, we show that DNA ligases, which seal single-stranded nicks in double-stranded DNA (dsDNA), also operate on DNA-gold monoconjugates, structures in which a single strand of thiolated DNA is bound to a gold nanocrystal. Previous work has demonstrated that large networks may be formed through ligation of nanoparticles whose surfaces are conjugated with (1) Alivisatos, A. P.; Johnsson, K. P.; Peng, X. G.; Wilson, T. E.; Loweth, C. J.; Bruchez, M. P.; Schultz, P. G. Nature (London, U.K.) 1996, 382, 609–611. (2) Katz, E.; Willner, I. Angew. Chem., Int. Ed. 2004, 43, 6042–6108. (3) Niemeyer, C. M. Angew. Chem., Int. Ed. 2001, 40, 4128–4158. (4) Voet, D.; Voet, J. G. Biochemistry, 2nd ed.; John Wiley and Sons, Inc.: New York, 1995. (5) Kanaras, A. G.; Wang, Z. X.; Brust, M.; Cosstick, R.; Bates, A. D. Small 2007, 3, 590–594. (6) Kanaras, A. G.; Wang, Z. X.; Bates, A. D.; Cosstick, R.; Brust, M. Angew. Chem., Int. Ed. 2003, 42, 191–194. (7) Pena, S. R. N.; Raina, S.; Goodrich, G. P.; Fedoroff, N. V.; Keating, C. D. J. Am. Chem. Soc. 2002, 124, 7314–7323. 9598

9

J. AM. CHEM. SOC. 2008, 130, 9598–9605

large numbers of DNA strands.8,9 In this work, ligation causes the formation of a covalent bond between discrete nanoparticleDNA conjugates, creating a new type of building block for selfassembly. These new nanoassembly components offer a number of advantages. In these building blocks, multiple gold nanoparticles are linked by single-stranded DNA (ssDNA) (Scheme 1) and thus retain the sequence-specific targeting ability of the ssDNA linker. This contrasts with previous experiments, in which DNA-gold dimers were formed by hybridization of two DNA-gold monoconjugates10 and thus are intrinsically linked by dsDNA. Additionally, since particles within these new ligated building blocks are covalently linked, the particles will remain connected under conditions such as high temperature or low salt concentrations that would cause disassembly in structures formed by base pairing. The ability to preform multiparticle components for later assembly also is a powerful asset in that it allows preformed components to be purified individually, which may ultimately facilitate assembly into complex structures such as nanoscale devices. The capability to operate with DNA-active enzymes on discrete DNA-nanocrystal conjugates also opens the door to the use of enzymatic chain reaction technology11,12 to apply amplification in nanomaterial construction. Similar to the manner in which PCR can be used to amplify tiny amounts of a sequence of DNA bases, we used a ligase chain reaction (LCR)12 to amplify the quantity of ligated DNA-gold dimer structures starting from small amounts of template. Since the ligation (8) Kanaras, A. G.; Wang, Z. X.; Hussain, I.; Brust, M.; Cosstick, R.; Bates, A. D. Small 2007, 3, 67–70. (9) Xu, X. Y.; Rosi, N. L.; Wang, Y. H.; Huo, F. W.; Mirkin, C. A. J. Am. Chem. Soc. 2006, 128, 9286–9287. (10) Loweth, C. J.; Caldwell, W. B.; Peng, X. G.; Alivisatos, A. P.; Schultz, P. G. Angew. Chem., Int. Ed. 1999, 38, 1808–1812. (11) Saiki, R. K.; Gelfand, D. H.; Stoffel, S.; Scharf, S. J.; Higuchi, R.; Horn, G. T.; Mullis, K. B.; Erlich, H. A. Science (Washington, DC, U.S.) 1988, 239, 487–491. (12) Barany, F. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 189–193. 10.1021/ja8026746 CCC: $40.75  2008 American Chemical Society

Enzymatic Ligation of Nanoparticle Conjugates

ARTICLES

Scheme 1. Overview of Ligation Processa

a Nanoparticle-DNA monoconjugates are hybridized to a ssDNA template and enzymatically ligated to create gold nanoparticle dimers. The template strand is then removed, leaving two nanoparticles connected by ssDNA.

process we describe also effectively polymerizes the nanoparticle conjugates to make short chains of nanoparticles, this can be considered to be roughly equivalent to PCR for nanoparticless here, we demonstrate formation of the shortest possible chain (two particles). Optimization of the amplification process could potentially allow construction of longer chains consisting of an arbitrary sequence of nanoparticles, which could then act as the template for formation of a complementary sequence of nanoparticle bioconjugates to engineer properties such as plasmonic coupling13 between pairs of nanoparticles on the chain. The efficiency of such a process would be limited relative to true PCR since it relies on the action of a DNA ligase, which is typically less efficient than a DNA polymerase. However, we demonstrate that our modified ligase chain reaction protocol can produce quantities of DNA-gold structures practical for selfassembly experiments. Ligation of discrete gold nanoparticle monoconjugates bearing a single strand of DNA offers substantial rewards in terms of control over self-assembly but levies additional requirements on the substrates involved. DNA used for conjugation must be highly purified, as the ligation process depends critically upon the matching of 3′ and 5′ ends of DNA at the ligation site. The nanoparticle surface must be thoroughly passivated (in this case, by thiolated poly(ethylene glycol) ligands) to prevent aggregation in enzymatic buffers containing divalent cations. Further, monoconjugates purified via gel electrophoresis14 are not ideal substrates as the samples may contain agarose impurities that are potent inhibitors of ligases.15 We used conjugates purified by anion exchange high-performance liquid chromatography (AE-HPLC).16 Results and Discussion

We first demonstrate successful hybridization and ligation as shown in Scheme 1 by analyzing reaction products in an agarose gel. Template removal was confirmed by small-angle (13) Sonnichsen, C.; Reinhard, B. M.; Liphardt, J.; Alivisatos, A. P. Nat. Biotechnol. 2005, 23, 741–745. (14) Zanchet, D.; Micheel, C. M.; Parak, W. J.; Gerion, D.; Alivisatos, A. P. Nano Lett. 2001, 1, 32–35. (15) Sambrook, J.; Russell, D. W. In Molecular Cloning: A Laboratory Manual, 3rd ed.; Argentine, J., Ed.; Cold Spring Harbor Laboratory Press: New York, 2001; Vol. 1. (16) Weiss, J. Handbook of Ion Chromatography; Wiley: New York, 2004; Vol. 1.

X-ray scattering (SAXS) experiments. Multinanoparticle ssDNA building blocks then were self-assembled into complex nanoscale structures that were visualized by transmission electron microscopy (TEM). Finally, the ability to use an enzymatic exponential amplification method was explored as a means to amplify nanoparticle-DNA building block products. Enzymatic Ligation Analyzed by Electrophoresis. Gold nanoparticles functionalized with a single strand of thiolated DNA were isolated using an AE-HPLC protocol previously described.17 Enzymatic ligation using the T4 and Taq DNA ligase enzymes requires nicked dsDNA, with the 5′ side of the nick phosphorylated.12,18,19 For these experiments, gold nanoparticles were conjugated to 30 base sequences A and B, where A was purchased with a commercially available thiol modifier at its 3′ terminus and was 5′-phosphorylated, and B was 5′-thiolated (Scheme 1). Hybridization of conjugates A and B to the 54 base template strand nAcBcn, in which the central 30 base AcBc sequence is complementary to conjugate sequences A and B, and the 12 bases on each end labeled n do not hybridize, leaving a toehold for later removal of the template. Hybridization of conjugates and template created a nicked double-stranded hybrid, in which the 5′ side of the nick was the phosphorylated end of conjugate A. Addition of T4 DNA ligase and incubation at 16 °C for 30 min resulted in the formation of a covalent bond between the adjacent ends of conjugates A and B, creating a ligated dimer. Following ligation, the template strand was removed by the addition of a DNA strand ncABnc complementary to the template. When DNA-gold conjugate structures are analyzed by agarose gel electrophoresis, they give discrete bands corresponding to structures with increasing numbers of DNA strands and nanoparticles.20 We used electrophoresis to confirm dimer hybridization and enzymatic ligation. In an agarose gel, constructs travel at a rate that depends upon both their size and their charge, with smaller structures passing more readily through the pores in the gel. Thus, in the electrophoretic analysis (17) Claridge, S. A.; Liang, H. W.; Basu, S. R.; Frechet, J. M. J.; Alivisatos, A. P. Nano Lett. 2008, 8, 1202–1206. (18) Takahashi, M.; Yamaguchi, E.; Uchida, T. J. Biol. Chem. 1984, 259, 41–47. (19) Engler, M. J.; Richardson, C. C. The Enzymes; Academic Press: San Diego, 1982; Vol. 5. (20) Zanchet, D.; Micheel, C. M.; Parak, W. J.; Gerion, D.; Williams, S. C.; Alivisatos, A. P. J. Phys. Chem. B 2002, 106, 11758–11763. J. AM. CHEM. SOC.

9

VOL. 130, NO. 29, 2008

9599

ARTICLES

Claridge et al.

Figure 1. Electrophoretic analysis of monoconjugate ligation efficiency. Lanes: (1) monoconjugate B, (2) A and B plus ssDNA template, (3) A and B plus ssDNA template and T4 DNA ligase, (4) A and B plus ssDNA template and ssDNA complementary to template, (5) A, B, ssDNA template and ligase, with subsequent addition of ssDNA complementary to template.

(Figure 1), monoconjugates (lane 1) such as A and B migrate more quickly than hybridized nicked dimers (lane 2). Ligated dimers (Figure 1, lane 3) migrate at a similar rate to nicked dimers (Figure 1, lane 2) since the ligation process does not change the charge of the construct and changes its size only insofar as the increased structural rigidity hinders progress through the porous gel. When unligated dimers such as those shown in lane 2 are mixed with the DNA strand complementary to the template strand, heated, and then slowly cooled, the template and its complement form a more stable hybrid than the template and the two conjugates. Thus, when the reaction products are analyzed in the gel (Figure 1, lane 4), the primary visible product is monoconjugate rather than dimer. However, when ligated dimers are mixed with the template complement strand, heated, and then cooled, the majority of the particles (71% by optical density analysis) still migrate as dimers in the gel (Figure 1, lane 5). While it is not clear, based on gel analysis alone, as to whether or not the template strand has been removed by its complement, lane 5 in Figure 1 demonstrates successful ligation. Analysis of Ligation and Template Removal by SAXS. To demonstrate that the template strand has indeed been removed through the addition of its complement, leaving a ssDNAnanoparticle dimer, the ligated dimers were analyzed by SAXS. The relatively high electronic density of the gold nanoparticles and low electronic density of DNA resulted in data that reflect the distribution of distances between the gold nanoparticles, without interference from the DNA between them. Previous studies21,22 elucidated the distances between nanoparticles connected by dsDNA, demonstrating that the interparticle distance reflects the length of the double-helical DNA linker. SAXS was used to observe the structural difference between ligated nanoparticle dimers bound to the template and ligated (21) Park, S. J.; Lazarides, A. A.; Storhoff, J. J.; Pesce, L.; Mirkin, C. A. J. Phys. Chem. B 2004, 108, 12375–12380. (22) Reinhard, B. M.; Sheikholeslami, S.; Mastroianni, A.; Alivisatos, A. P.; Liphardt, J. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 2667–2672. 9600

J. AM. CHEM. SOC.

9

VOL. 130, NO. 29, 2008

Figure 2. (a) Schematic of gold nanoparticle dimers connected by a single strand of ssDNA. (b) SAXS analysis shows a decrease in interparticle distance after removal of the template strand.

dimers from which the template was removed (Figure 2a, similar to Figure 1, lanes 3 and 5). In both structures, the nanoparticles are separated by a 60 base DNA linker, but in the structure on the left (Figure 2a), the central 30 bases of the linker are hybridized to the template, creating an ∼10 nm double helix flanked by 15 base lengths of flexible ssDNA. In the structure on the right in Figure 2a, after the template strand has been removed by addition of the template complement, the nanoparticles are separated by 60 bases of ssDNA. Since the persistence length of ssDNA (∼2 nm) is much shorter than that of dsDNA (50 nm),23 removal of the template is expected to lead to smaller interparticle distances. Figure 2b shows the effect of adding the template complement strand. When no template complement is added (red trace in Figure 2b), the SAXS pair distribution function shows a peak at 21.2 nm. When 2 equiv of template complement is added and the sample is allowed to equilibrate for 24 h (blue trace in Figure 2b), a large peak appears at ∼11.7 nm, with a small residual peak corresponding to structures from which the template has not been removed. What appears to be a fine structure in the residual peak is actually noise from Fourier transformation, which is commonly observed for lower intensity signals in this type of data. Since the peaks at 11.7 and 21.1 nm are indicative of centerto-center distances between nanoparticles, this represents an endto-end distance of ∼16 nm for the DNA linker in the first case and 6 nm in the second case. Thus, although there is little difference in the migration of the two structures in an agarose gel, distinctive characteristics in the pair distribution functions allow them to be differentiated unambiguously. Self-Assembly of Ligated Nanoparticle Dimer and Trimer Structures. As a demonstration that ligated ssDNA dimers and

trimers retain the targeting ability of their DNA linker, ligated dimers and trimers were synthesized and assembled to form

Enzymatic Ligation of Nanoparticle Conjugates

ARTICLES

Scheme 2. Overview of Self-Assembly Processa

a Nanoparticle dimers and trimers were created by enzymatic ligation of mono- or diconjugates as shown. Hybridization of complementary dimers or trimers led to nanoparticle tetramers or hexamers.

nanoparticle tetramers and hexamers (Scheme 2). Tetramers were formed by hybridization of two complementary ligated dimers. To synthesize hexameric structures, nanoparticle diconjugates were first specifically ligated to two monoconjugates to form a symmetric strand of three nanoparticles connected by ssDNA. Two such complementary strands were synthesized and subsequently hybridized to yield a hexameric nanoparticle structure comprising three pairs of nanoparticles. Figure 3a shows transmission electron micrographs of ligated nanoparticle dimers that were incubated with template complement DNA to remove the ligation template strand. As a result, the two nanoparticles are attached to a length of ssDNA. Because of the decreased persistence length of ssDNA relative to dsDNA, the center-to-center interparticle distance varies widely, with an average of 20.9 ( 14.0 nm. The behavior of nanoparticle-DNA constructs dried on TEM grids differs markedly from their behavior in solution (Figure 2b), where the center-to-center distance of the same structures is 11.7 ( 2.8 nm. We presume that forces exerted on the sample while drying on the TEM grid override the entropic penalties associated with ssDNA stretching, resulting in larger interparticle distances as well as a larger distribution in the distances. The four particle pairs shown in Figure 3a demonstrate the extremes of both stretching (top and left pairs) and collapse (right pair), as well as the average case (bottom pair). In Figure 3b, complementary ligated dimers were hybridized to form nanoparticle tetramers, in which the increased structural rigidity provided by hybridization is evident. The majority of tetramers consist of two pairs of particles. If the long axis of the tetramer is taken to be the distance between the midpoints of the two pairs, the distribution is found to be 18.6 ( 2.6 nm, in agreement with the expected 20 nm length of a 60 base DNA double helix. The much smaller standard deviation reflects the increased rigidity of the structure. The ligation strategy also can be extended to create more complex nanoscale building blocks. In Figure 3c, hybridization of two three-particle strands creates nanoparticle hexamers consisting of three pairs of particles. Because of the random distribution of the two DNA strands on the nanoparticle surface in the diconjugate, the observed hexamer structure morphology is not necessarily linear as depicted in the schematic representation. Visual interpretation of these larger particle patterns can be complicated by the variety of angles and ambiguities that arise when two groupings are positioned close together on the grid. For clarity, in Figure 3c, we present a somewhat narrower field of view in which all of the particle groupings are wellseparated. Exponential Amplification of Ligated Nanoparticle Dimers. 11

Polymerase chain reaction (PCR) and ligase chain reaction (LCR)12 have found wide use as a means of amplifying small

Figure 3. Transmission electron micrographs of ssDNA dimers and trimers hybridized to form larger multiparticle structures: (a) ligated 60 base ssDNA dimers, (b) complementary ligated dimers hybridized to form nanoparticle tetramers, and (c) complementary ligated trimers hybridized to form nanoparticle hexamers.

amounts of DNA. While PCR has found wider application due to its robustness and ease of use, LCR also is applied in certain systems as a means of detecting single-nucleotide sequence changes.12,24 Success in such LCR experiments relies on a careful choice of reaction temperatures:12 too low a temperature results in reduced selectivity for the desired single-base substitution, while too high a temperature results in a lower amplification efficiency due to DNA melting. We demonstrate the use of nanoparticle monoconjugates as substrates for a ligase chain reaction, amplifying the amount of DNA-gold dimer product (23) Bustamante, C.; Bryant, Z.; Smith, S. B. Nature (London, U.K.) 2003, 421, 423–427. (24) Abravaya, K.; Carrino, J. J.; Muldoon, S.; Lee, H. H. Nucleic Acids Res. 1995, 23, 675–682. J. AM. CHEM. SOC.

9

VOL. 130, NO. 29, 2008

9601

ARTICLES

Claridge et al.

Scheme 3. Amplification of DNA-Gold Nanostructures via LCR

formed through repeated thermal cycling in the presence of Taq DNA ligase. In these experiments, since our focus is on material construction rather than single-base substitution detection, selectivity constraints do not impact the choice of ligation temperature. Thus, we are able to perform ligation at a somewhat lower temperature (chosen as the minimal temperature for good activity of the thermophilic enzyme), resulting in more stable DNA hybridization and presumably therefore in higher reaction efficiencies than we might otherwise achieve. In this LCR (Scheme 3), two strands of DNA were first hybridized to a cDNA template to form a nicked double helix. The reaction mixture then was incubated at a moderate temperature (45 °C) to allow the Taq DNA ligase enzyme to form a covalent bond at the nick site, and the reaction mixture subsequently was heated to 80 °C to melt the template from the ligated dimer. Since this exceeds the temperature range in which the thiol-gold bond is stable,25 trithiol terminated DNA is used in synthesizing conjugates to create more strongly bound structures. Following the high-temperature step, the temperature was lowered to 45 °C, which allows the DNA to rehybridize. In the reaction mixture, conjugates and template halves initially were present in large excess. Thus, although it is possible for a ligated dimer to rehybridize with a full template, it is statistically more likely that the template will hybridize with conjugate monomers and that the ligated dimer will hybridize with template halves. After the second ligation cycle, assuming 100% ligation efficiency, there will be twice as much ligated dimer. Repeated thermal cycling then can lead to exponential growth in the amount of ligated dimer product, as 2n. The amount of dimer present can be monitored by gel electrophoresis (Figure 4a) and plotted against the number of thermal cycles (Figure 4b) to demonstrate exponential growth. In practice, ligation efficiency is